Filling subnational sanitation data gaps in low- and middle-income countries

For billions of people, the simple act of using a toilet is not a given. Access to safe sanitation is a fundamental human right, yet nearly 1.5 billion people around the world still lack access to basic services. This gap is not just a statistic; it is a driver of disease, child stunting, and preventable death. To fix the problem, governments and aid organizations need to know exactly where the need is greatest. They need to see the map of sanitation, not just for entire countries, but for the specific regions within them. However, in many low- and middle-income nations, this detailed map is missing. The data that does exist often comes from household surveys that are expensive to conduct, take years to complete, and leave vast areas of the country unmeasured. Without up-to-date, granular information, it is difficult to direct resources to the communities that need them most.

A team of researchers has now filled in these missing pieces for 130 countries using a new approach that combines old survey data with modern satellite technology. By analyzing information from 1.8 million households and layering it with environmental and economic data, they have created a detailed picture of sanitation coverage for 2024. Their work reveals a stark reality: the poorest and most fragile regions, particularly in sub-Saharan Africa, are falling furthest behind. While the study provides the clearest view yet of where sanitation services are lacking, it also highlights the limits of what we can know without more direct data, offering a transparent guide for where the next surveys should go.

The researchers faced a significant challenge: they needed to predict sanitation conditions in areas where no recent surveys existed. Traditional methods of guessing these numbers often fail when data is scarce, tending to overfit the limited information they have or producing unreliable results for new areas. To solve this, the team turned to a sophisticated type of artificial intelligence called a Tabular Prior-Fitted Network. Think of this model as a highly trained expert who has studied millions of synthetic examples of how different factors—like population density, rainfall, and national wealth—relate to sanitation access. This allows the model to make accurate predictions even when it only has a small amount of real-world data to work with, a situation common in the world's most data-scarce regions.

The team fed the model data from the most recent household surveys conducted between 2015 and 2024, covering 1.8 million homes across 89 countries. They combined this with a vast array of geospatial data, including satellite images of built-up areas, nighttime lights, soil types, and climate patterns. The model learned to recognize the patterns that link these environmental and economic factors to whether a community has access to a basic toilet or is forced to practice open defecation. Once trained, the model was used to predict sanitation coverage for every first-level administrative region in 130 low- and middle-income countries, effectively creating a global map of sanitation access for the year 2024.

The results paint a picture of deep inequality. In upper-middle-income countries, the median coverage for basic sanitation is 94 percent, meaning almost everyone has access to a toilet. In contrast, in low-income countries, the median coverage drops to just 36 percent. The situation is even more dire in sub-Saharan Africa, where the median coverage is 39 percent, and in fragile, conflict-affected settings, where it falls to 48 percent. The reverse is true for open defecation, which involves disposing of human waste in open fields or waterways. In upper-middle-income countries, fewer than 1 percent of people practice open defecation, but in low-income countries, that figure rises to 8 percent, with some regions seeing rates as high as 73 percent.

The study also mapped the uncertainty of these predictions, showing where the data is most reliable and where it is a guess. The model is most confident in low- and middle-income regions of Europe and Northern America, Eastern and South-Eastern Asia, and Northern Africa and Western Asia, where recent surveys and national estimates exist. However, in low-income and fragile settings, particularly in sub-Saharan Africa, the uncertainty is highest. This is not a failure of the model, but a reflection of the reality that these areas lack the ground-level data needed to be certain. The researchers identified four specific countries—Republic of Congo, Eritrea, Nicaragua, and North Korea—as top priorities for new data collection, as they lack recent surveys and have large populations living in conditions of high uncertainty.

One of the most important findings is that the inequalities exist not just between countries, but within them. Even in nations with moderate overall coverage, there are regions where sanitation access is nearly non-existent. The model successfully identified these pockets of deprivation, though it tended to smooth out the sharpest differences, likely underestimating the extreme gaps in the worst-off areas. This suggests that while the map is a powerful tool for identifying broad trends, local data remains essential for targeting the very worst-off communities. The researchers also noted that their model sometimes underestimated the prevalence of open defecation, a known issue in surveys where people may be reluctant to admit to the practice.

The study confirms that progress toward the global goal of ending open defecation by 2030 is dangerously slow. To reach that target, the speed of progress would need to triple compared to the last decade. The new maps show that without a massive increase in investment and targeted action, the most vulnerable populations will be left behind. By providing a transparent, uncertainty-aware approach to filling data gaps, this work offers a clear path forward. It tells us not only where the problem is, but also where our knowledge is incomplete, guiding the next generation of surveys and investments to the places where they are needed most.